Gold Price Volatility Characteristics Of The Conversion Model Based On Markov Mechanism

Many economic variables usually experience sudden changes in some time. Histrionic changes would show in their time series process. The remarkable change in a time series is sometimes regarded as a switch in the regression equation from one regime to another. As for the complicated time series, we can create models in sections, if the specific time when the regime switches can be obtained. In many cases, however, when the switch happens and when the model parameters change is unknown. Markov regime-switching model takes the regime-switch as a random endogenous variable, which enable to depict the remarkable change time series in one model.Gold price has experienced great changes in the past decades of years. In this paper, we take Markov-switching model to analyze the volatility of gold price. We choose London fix gold monthly return time series from January,1980 to December,2009. First, we make a nonlinear test and a regime-switch test. Then we create a three states and two-step lag Markov regime-switching model. According to the result of estimation and the analysis of smooth probabilities, we find that gold price fluctuation exhibits three different kinds of moving situation:rising mildly, dropping sharply and rising sharply. The average holding period of each situation is 4.3 months,2.2 months and 34 months. Situations switch to each other in certain probability. At last, we have compared the estimation results of the Markov-switching model and AR(2) model. We can make a conclusion that the Markov-switching model is better than AR(2) model in describing the change of gold price.